1
Faculty Economics and Business Bachelor’s thesis
The effect of the enlargement of the
European union on its free trade agreements
By Robert Rijneker
Bsc Economics and Business
Specialization: Economics and Finance
Academic year: 2017-2018, semester 1
Student number: 10791094
Supervisor: Ioana Neamtu
2
Statement of Originality
This document is written by Student Robert Rijneker who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Acknowledgements
First of all I would like to thank my supervisor, Ioana Neamtu, for helping me through the process. She always responded quickly with good critical feedback and useful hints. She helped me to keep sight of the big picture. Furthermore, I would like to thank my friends for studying with me and for making the days in the library even more entertaining. Last, I would like to thank my parents for believing in me and for the amazing support they have been for me the last months.
4
Abstract
The aim of this thesis is to analyze whether there is a difference between the effect of the Free Trade Agreements of the European Union on trade before and after the enlargement in 2004. The effect of FTA’s on trade are estimated with a theoretically-motivated gravity model using panel data with fixed time effects and fixed cross-country effects. Quarterly data on 8 countries is used. For four countries, the FTA went into force before 2004 and for four countries after 2004. The qualitative comparison between the two betas that estimated the effect of the FTA’s indicate that the enlargement of the EU had a positive effect on its FTA’s. Specifically, the FTA’s effect on trade was not found significant for the countries with the FTA’s set before the enlargement. The FTA’s of the countries with the FTA’s set after the enlargement had a positive effect on trade of 12.55%.
5 Table of Contents Introduction ... 6 Literature review ... 8 Context ... 8 Gravity model ... 9 Estimation methodology ... 11
Methodology & empirical analysis ... 13
Methodology ... 13
Hypothesis ... 15
Data ... 15
Main Results ... 17
Analysis ... 18
Discussion & Conclusion ... 19
Discussion ... 19
The impact of fixed effects ... 20
Conclusion ... 21
6
Introduction
The European Union (EU) is an advantageous trading partner for multiple countries. At the moment, the EU is expanding their area of free trade and has over thirty Free Trade
Agreements (FTA’s) signed and is negotiating with nine more countries (European
Commission, 2017). The connectivity between two trading partners will rise, due to the fall of trade barriers and this will create a larger amount of trade.
There are also opponents towards the FTA’s of the EU, especially towards the Deep and Comprehensive Free Trade Agreement with Ukraine. On April 6th, 2016, there was a referendum in the Netherlands on the Association Agreement between the EU and Ukraine. 61.1% of the voters was against this DCFTA (Kiesraad, 2016).
It should be well examined whether FTA’s of the EU have an effect on trade and if these effects are rising or decreasing during the past years. Until now, the EU has been expanding since its foundation. The major enlargement of the EU took place in 2004 (Breuss, 2009). I chose this important point in the history of the EU to investigate whether the effects of the FTA’s of the EU on trade after 2004 are different from before 2004. The effects of the FTA’s are namely expected to increase as the Free Trade Area is increasing. Therefore, the effect of the enlargement of the EU in 2004 with ten states on its FTA’s is examined in this thesis.
Since there are nine negotiations for FTA’s outstanding with the EU, this thesis aims to analyze whether FTA’s are becoming more or less relevant. This is important because during the negotiations the European Union can emphasize the positive effects of the upcoming agreements.
To summarize, in this context my research question is to analyze whether there is a difference between the effect of the Free Trade Agreements of the European Union on trade before and after the enlargement in 2004.
This question will be answered by assessing the FTA’s effect of 8 different countries. For four countries, the FTA went into force before 2004 and for four countries after 2004. The effects of the FTA’s of Albania, Georgia, Serbia and South Korea is compared to the effects of the FTA’s of Israel, Mexico, Morocco and Armenia. Ukraine is not included in the dataset, because this FTA is set too close to present to analyze its effect. The gravity model is the main model to analyze the effect of FTA’s on trade and will therefore be used in this thesis as well (Deardorff, 1995; Rose, 2004). The gravity model is augmented with
7 to control for external events.
FTA’s lead to free trade areas and these have multiple advantages for their members. More economies of scale, better use of comparative advantages and higher exports due to lower tariffs. When the free trade area becomes larger, these positive effects on their bilateral trade are expected to increase as well (Krugman et al., 2009). Therefore, I expect the effect of the FTA’s of the EU on trade to rise after the enlargement in 2004.
In order to analyze this hypothesis, first the literature will be reviewed. The gravity model has to be specified carefully in order to get unbiased results (Baier & Bergstrand, 2007). Therefore, the impact of adding fixed effects to the regression will be discussed. A panel regression including fixed effects allows to isolate the effect of the FTA’s on trade and at last a conclusion can be made due to analyzing the data (Cheng & Wall, 2005). The specification of the gravity model has an effect on its results, so this thesis tries to contribute to the academic debate on how to use the gravity model.
After this introduction chapter 2 consists of the theoretical context of the FTA’s and its background. Also, the gravity model is explained with its methodology. Chapter 3 describes the methodology and empirical analysis. Chapter 4 discusses the results found in this thesis and at last a conclusion will be made.
8
Literature review Context
One of the first theories on economic integration came from Bela Balassa (1961). Balassa divides economic integration into five different stages. Each upcoming stage has a deeper level of economic integration. The agreement with the lowest level of economic integration is the Preferential Trade Agreement. This is a trade pact that reduces tariffs on exports and imports between member countries. However, this is not a stage in the economic integration theory by Balassa. The first stage of economic integration is called a free trade area. This is also the most common type of economic integration. The main goal of a free trade area is to develop economies of scale and comparative advantages, which promotes economic
efficiency. Each member of the free trade area eliminates their import and export tariff for the other members. At the same time, the members keep their right to have individual trade policies with non-members (Balassa, 1961).
The second stage of economic integration is a Customs Union. A Customs Union eliminates trade tariffs for its members just as the free trade area. However, the Customs Unions has a common trade regime as well. This means that the member countries set common external trade tariffs for non-member countries. The next stage of economic integration is a Common Market. Capital and services are free to move within a Common Market, but each member country has its own regulations. The fourth stage of economic integration is an Economic Union. It has the same characteristics as a Common Market, but an Economic Union has harmonized fiscal and monetary policies as well. An option for an Economic Union is to go to a monetary union. This means the Economic Union agrees to use a common currency. The last stage of the economic integration theory by Bela Balassa is a Political Union. This is the most advanced level of integration policy, since this implies a common government (Balassa, 1961).
This thesis analyzes the effect of Free Trade Agreements. These Free Trade Agreements lead to the first stage of economic integration: a Free Trade Area. There are multiple advantages of free trade. Free Trade Agreements eliminate trade tariffs and therefore the products and services will be cheaper to import. This leads to increased exports for the member countries (Krugman et al., 2009). By eliminating trade barriers, consumption will shift from high-cost producers in the home country to low-cost producers in the member country. This way, free trade enables countries to specialize in products in which they have a comparative advantage (Dornbusch et al., 1977). Producers have a comparative advantage
9
when they can produce goods and services at a lower opportunity cost than other economic actors (Krugman et al., 2009). If member countries specialize in the products and services for which they have a comparative advantage and later trade these products and services, they will benefit. This theory of comparative advantage is founded by David Ricardo and first published in 1833 by William Whewell (Wood, 1991). Another advantage of free trade is called economies of scale. The area of distribution becomes larger, so there could be a decline in average costs due to the increase in production (Koshal, 1972). This is especially beneficial for industries with high fixed costs. The sales will increase and this will lead to lower average costs for consumers, which will lead to more sales (Wadhva, 1971). This way a vicious circle is created that creates welfare for member countries. Moreover, trade will shift from non-member foreign countries towards the member country. Free Trade Agreements will thus likely increase the bilateral trade between member countries.
The EU is according to the theory of economic integration by Balassa (1961) an Economic Union. This step towards an Economic Union took place on February 7th, 1992 in Maastricht. The Maastricht Treaty was the official document that took the EU from a
Common Market to an Economic Union (European Commission, 1992). The Euro was launched ten years later. 12 members of the EU began with the common currency. Every member state of the EU has to implement the Euro as their common currency if they meet the monetary standards. Only Denmark and Great-Britain negotiated for a position of exception. Nowadays, 22 countries have implemented the Euro (European Commission, 2018). This major step in integration of the EU is called the Economic and Monetary Union (EMU).
Another major step for the EU was the enlargement of the EU in 2004. Czech
Republic, Estonia, Cyprus, Latvia, Lithuania, Hungary, Malta, Poland, Slovenia and Slovakia joined the EU (Breuss, 2009). After the enlargement, the EU became the largest internal market of the world and accounted for 19% of the total world trade (European Commision, 2005).
Gravity model
The impact of the EU on intra-EU-trade has been studied over the past decades. The effect of FTA’s of the EU is mostly only studied individually (Soete & Van Hove, 2017). This is because both parties for which the FTA was set want to analyze the ex post effect of the FTA. These researches on specific FTA’s provide insight into the agreement-specific impact for these countries, but this approach lacks generalizability.
10
Abrams, 1980; Aitken, 1973; Bergstrand, 1985; Frankel et al., 1997). Baier & Bergstrand (2007) did not research FTA’s individually, but used a general approach. They studied every FTA and found that the effect on trade by FTA’s was underestimated for years. They found that an FTA on average will increase trade by 100% after ten years.
Soete & van Hove (2017) used a general approach as well. They wanted to analyze the ex post trade effects of the Economic Integration Agreements of the EU. They found that deeper economic integration like FTA’s and Customs Unions lead to more enhancing trade than Preferential Trade Agreements. In their research, they also did individual regressions for some countries used in this thesis. They found a statistically significant negative effect on trade for Albania and no significant effects on trade for Israel, Morocco and Mexico. However, as stated above, this approach lacks generalizability.
Furthermore, there is a little knowledge about the effects of the enlargement of the EU (Belke & Spies, 2008). There is a lack of information on the overall effect of FTA’s of the EU on trade, especially on the difference of this effect before and after the enlargement of the EU. In this thesis, the enlargement of 2004 will be discussed, since this is the major
enlargement of the EU. There are no findings yet on this particular topic.
We can measure the effect of an FTA using the gravity model, that originates at Jan Tinbergen (1962). The gravity model became widely used and perceived empirical
acknowledgements since its foundation. However, there were also critics about the gravity model, since it did not have a theoretical foundation on itself. Later this foundation became developed by Anderson (1979) and Bergstrand (1985), who derived the model from the models of monopolistic competition. Deardroff (1995) developed to the theoretical
foundation even further when he showed that the gravity model could be derived within the Ricardian and Heckscher-Ohlin frameworks.
Export is the dependent variable in the generalized gravity model of trade. The export, Xij, from country i to country j is a function of both countries their GDPs, their populations, their geographical distance and a set of dummies. For which Yi (Yj) is the GDP of the exporter (importer), Ni (Nj) indicates the population of the exporter (importer), Dij measures the distance between their economic center and E is the error term (Tinbergen, 1962).
The dependent and independent variables are measured in their logarithms. This is referred to as a log-log model. In the log-log model, the estimated betas are the elasticity of the dependent variable with respect to the independent variable in question.
11 𝛽1 = Δ𝑌 𝑌 Δ𝑋 𝑋 = 100 ∗ ( Δ𝑌 𝑌 ) 100 ∗ (Δ𝑋𝑋 ) = 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑐ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑦 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑐ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑥
This means that a 1% change in the independent variable will lead to a 𝑒 𝛽− 1 = Δ𝑌 change in the dependent variable (Stock & Watson, 2008).
The traditional variables used in the gravity model are sensible, economically and statistically significant and reasonably consistent across studies (Rose, 2004). The model is later praised for its high explanatory power (Anderson, 2011). The model fits the data on trade well and has been reliable since its existence. Therefore, in order to answer the research question, the effect of an FTA will be measured using a general approach of the gravity model.
Estimation methodology
Most plausible estimates for FTA’s on trade are made with a theoretically-motivated gravity model using panel data with fixed time effects and fixed cross-country effects (Baier & Bergstrand, 2007). The standard gravity model as described in previous paragraph is augmented with theoretically-motivated variables as described in the next chapter.
An omitted variable bias appears, when using cross-country data (Bayoumi &
Eichengreen, 1997; Matyas, 1997). The dummy variables will capture everything specific to the exporting or importing country, what is not captured by the included variables. This means that the countries’ heterogeneity is omitted and this bias will lead to unwanted results, since the dummy variables need to isolate the effect of the FTA on trade. With adding fixed effects to regression, the panel data controls for the individual heterogeneity. Moreover, panel data is not limited to one unit like time-series data and consists of multiple time periods unlike cross-sectional data (Cheng & Wall, 2005).
Micco et al. (2007) observed that the effect of an FTA on trade is systematically smaller when pair dummies are included. One explanation for this is that due to reversed causality, the results of the effects of an FTA on trade can be biased. Since the benefits of FTA’s, such as economies of scale, are greater when the countries in question trade more, it is conceivable that the countries in question had high trade flows before having an FTA. In short, unusually high trade flows lead to FTA’s and not vice versa, which causes an
endogeneity problem (Baier & Bergstrand, 2007). Therefore, time fixed effects and fixed cross-country effects are included in this regression model.
12
could still be unusually high trade flows between the countries in the dataset in the sample period (Micco et al., 2007). However, by keeping the sample period relatively short (13 years per researched FTA), this alleviates the problem.
Another solution to the endogeneity problem is to use instrumental variables (Hausman & Taylor, 1981). When we leave out the fixed effects, the explanatory dummy variable FTAijt is correlated with the error term. Two conditions must hold to use the method of instrumental variables. The instrument must be correlated with the endogenous
explanatory variables and cannot be correlated with the error term (Stock & Watson, 2008). This instrument allows us to identify the unobserved correlation and to see the true
correlation between the dependent and independent variable. IV-regression splits the explanatory variable in two parts. One part is correlated with the error term and one part is not. An estimation of the explanatory variable in the regression can be made by isolating the part that has no correlation with the error term. In reality, it is hard to find such an instrument (Cavallo & Franko, 2008). Since in this thesis 8 FTA’s are observed and because of the time constraint, the instrumental variable method is not used in this thesis. Being aware that by using panel data with fixed effects, the endogeneity issue will not be solved completely, it is still possible to produce a useful research.
The betas of the two regressions in this thesis cannot be tested against each other. The first beta consists namely the effect of the FTA’s on trade for the countries Mexico, Israel, Morocco and Armenia for the years 1995 to 2004. The second beta consists the effect of the FTA’s on trade for the countries Korea, Albania, Georgia and Serbia for the years 2004 to 2017. Both betas measure the effect of FTA’s, but they measure it for different countries along different time periods. Another way to measure trade creation is the ex post income elasticity of import demand (YEM) (Balassa, 1974). The YEM is calculated by measuring the average annual percent change in observed imports and dividing this by the average annual percent change in observed GDP. Both changes must be adjusted for inflation. A rise in the YEM indicates trade creation and a fall in the YEM indicates trade diversion (Appleyard & Field, 2010). However, for the YEM applies the same problem as for the betas. The outcome can be compared qualitatively, but it cannot be tested at a significance level. Moreover, the YEM only uses data on imports and GDP and it does not take any events or other variables into account. Therefore, this thesis uses a qualitative comparison between the two betas incorporating sensible economic intuitions.
13
(1)
Methodology & empirical analysis Methodology
In this thesis, the gravity model is used as illustrated in equation 1. The dependent variable will be EXPORTijt, which is the logarithm of shipments of goods from region i to region j at time t expressed in the currency of country i. Using the logarithm of nominal exports as the dependent variable is in line with most of the literature on the gravity model (e.g., McCallum, 1995; Yang & Martínez-Zarzoso, 2013; Bergstrand, 1985).
𝐸𝑋𝑃𝑂𝑅𝑇𝑖𝑗𝑡 = 𝛼 + 𝛽1𝐺𝐷𝑃𝑖𝑡+ 𝛽2𝐺𝐷𝑃𝑗𝑡+ 𝛽3𝐺𝐷𝑃𝐶𝐴𝑃𝑖𝑡+ 𝛽4𝐺𝐷𝑃𝐶𝐴𝑃𝑗𝑡+ 𝛽5𝐷𝑖𝑠𝑡𝑖𝑗+ 𝛿1𝐹𝑇𝐴𝑖𝑗𝑡+ 𝜂𝑖𝑗 + 𝜆𝑡+ 𝜀𝑖𝑗𝑡
The first two independent variables, GDPit and GDPjt, measure the logarithm of the nominal gross domestic production at time t in the local currency for country i and j, respectively. Nominal GDP is chosen, so that the dependent and independent variables are defined in the same way.
The variables GDPCAPit and GDPCAPjt are added to the regression for country i and j, respectively to control the previous two independent variables by their population size. Both variables signify the nominal gross domestic production divided by the population size
and are measured in their logarithms.
In addition, the dummy variable is added for analyzing whether the enlargement of the EU in 2004 influenced the effectiveness of the FTA’s of the EU. In the first regression, the dummy variable FTAijt is used to determine if the country had a Free Trade Agreement at time t with the EU that is set before 2004. The dummy variable takes the value “1” if it exists and “0” otherwise. In the second regression, the dummy variable FTAijt does the same, but for a Free Trade Agreement that is set after 2004 with the EU.
Moreover, the gravity model includes fixed effects. ηij is the country-pair fixed effect between country i and country j. λt is the time fixed effect at time t. Further explanation on these fixed effects is given in the previous paragraph.
The last variable in the regression is the error term. The error term is granted to be clustered within country-pairs. Since a panel data approach is used, clustered standard errors are used rather than robust standard errors, because robust standard errors can give
inconsistent standard errors (Stock & Watson, 2008).
The standard gravity model adds a variable for controlling the geographical distance as well. This distance will imply the distance between the two economical capitals of the two
14
areas in question. However, this geographical variable is unnecessary when using country-pair fixed effects, since the data on geographical distance is the same across time and is country-specific (Cheng & Wall, 2005).
Trade is of different importance along countries. Therefore, weights are added to the regression. The weights are based on the exports of the country towards the EU divided by their GDP. The year for which the exports and GDP are measured is the year in which the FTA was set with the EU. Year rather than quarter is chosen to control for seasonal effects.
Table 1: Weights for the countries with an FTA set before the enlargement
*Source: IMF Direction of Trade Statistics (DOTS) **Source: International Financial Statistics (IFS)
Table 2: Weights for the countries with an FTA set after the enlargement
*Source: IMF Direction of Trade Statistics (DOTS) **Source: International Financial Statistics (IFS)
Countries with an FTA set before the enlargement
Mexico Israel Morocco Armenia
Export to the EU* (in millions of dollars)
5484 9214 5416 108
Gross Domestic Product** (in billions of dollars)
683.62 132.43 38.86 1.92
Weight 0.029335 0.254476 0.50964 0.206549
Countries with an FTA set after the enlargement
Korea Albania Georgia Serbia
Export to the EU* (in millions of dollars)
53522.41 949.088 621.206 9259.311
Gross Domestic Product** (in billions of dollars)
1202.24 12.04 16.51 44.21
15
Hypothesis
The original gravity model as described above tells that the dependent variable, export, is positively related to the GDP (Tinbergen, 1962). So, the exports are expected to increase as the GDP increases. The variable GDP is included in the model to control for this, such that the dummy variable that estimates the effect of the FTA can be isolated. This thesis is
interested in the connectivity that arises due to the FTA. The import of the home country will shift from non-member countries towards member countries. Consumption will shift from high-cost producers in the home country towards low-cost producers in the foreign country (Krugman et al., 2009). Moreover, countries will specialize in the goods in which they have a comparative advantage and due to economies of scale the exports will rise (Dornbusch et al, 1977). These positive effects are expected to increase as the free trade area becomes larger. The free trade area for which the countries set an FTA with expanded from 15 to 25 countries in 2004. Therefore, I expect the positive effects of the FTA’s on trade to rise after the
enlargement of the EU.
H0: ßuntill2004 = ßafter2004
H1: ßuntill2004 < ßafter2004
Data
In order to investigate the effect of the enlargement of the EU in 2004 on its FTA’s, quarterly data on 8 countries is used. For four countries, the FTA went into force before 2004 and for four countries after 2004.
I chose Albania, Georgia, Serbia and South Korea, because for these countries data is available for four years before and after the FTA was set.
I chose Israel, Mexico, Morocco and Armenia, because this data is most relevant. These FTA’s were set close to 2004 and enough data (at least four years) is available to investigate the effect of the FTA before the enlargement of the EU took place. Furthermore, their populations and GDP’s are roughly similar to the populations and GDP’s of Albania, Georgia, Serbia and South Korea.
Quarterly data on bilateral trade is gathered from the IMF Direction of Trade Statistics (DOTS). Quarterly GDP data is obtained from International Financial Statistics (IFS).
Population data came from the World Bank Databank. For the data on populations, only yearly data was available. This will lead to inexact numbers for the GDP per capita.
16
used do not differ more than a half percent between years, which means that for quarterly data large deviations are next to impossible. Therefore, the imprecise numbers of the GDP per capita will not lead to biased results.
I retrieved the data for Albania, Georgia, Serbia and South Korea on GDP, bilateral trade and population between July 1st, 2004 and July 1st, 2017. I gathered data for Israel, Mexico, Morocco and Armenia on GDP, bilateral trade and population between January 1st, 1995 and April 1st, 2004. The major enlargement of the EU took place on May 1st, 2004.
I used data from the European Commission (2017) for the date when the enlargement took place and for the dates when the FTA’s officially went into force. One exception is South Korea. According to the European Commission (2017) the FTA went into force in December 2015. However, the FTA was provisionally applied in July 2011. Using July 2011 as date is for which the FTA of South Korea went into force is in line with the existing literature (Soete & van Hove, 2017).
17 Main Results Table 3: Results (1) Baseline before enlargement (2) No FE before enlargement (3) Baseline after enlargement (4) No FE after enlargement lGDP -.9686* -.04574 .03134 .1312* (.3208) (.1493) (.0434) (.0423) lGDPEU 21.2274* 32.7954 15.9639* -47.2748 (5.0399) (28.5591) (4.3476) (26.7841) lGDPCAP 1.2229* -.8837* -.0100 .1008* (.3091) (.08729) (.1324) (.01704) lGDPCAPEU -20.6076* -27.4804 -15.0297* 47.0794 (5.3619) (30.4473) (4.5361) -28.3140 FTA .02839 -.8599* .1182* .8812* (.0698) (.3889) (.0543) (.3414) _cons -389.1233* -611.2278 -198.3102* 625.8722 (99.6771) (569.0407) (57.9777) (358.3853)
Country-pair FE Yes No Yes No
Time FE Yes No Yes No
R2 0.6807 0.4929 0.6253 0.2756
Clustered standard errors in parentheses
18
Analysis
This thesis aims to analyze whether there is a difference between the effects of FTA’s on trade before and after the enlargement of the EU, so the most important estimate in the model is the estimate of FTAijt. This variable estimates the effect of the FTA’s on export. For the countries with the FTA set before 2004, this variable is estimated at 0.0284. However, this variable is not statistically significant. The standard error of the variable is 0.0698, which results in a P-value of 0.685. This means that at a significance level of 5%, we cannot accept that the variable is statistically significant different than zero. This implies that for the Armenia, Mexico, Morocco and Israel we find no statistical evidence that there was any effect of the FTA’s on trade. For the countries with the FTA set after 2004, the variable FTAijt is estimated at 0.1182. The standard error of the variable is 0.0543. The P-value of the variable is 0.031, which means that the variable is statistically significant at a significance level of 5%. This implies that we can accept the alternative hypothesis that the variable is significantly different than zero. This implies that the FTA’s set in Korea, Albania, Serbia and Georgia contributed to a rise in exports of 12.55%1.
The outcomes for the variable lGDP are remarkable, since in the regression with the countries with the FTA set before 2004, the variable is significantly lower than zero. This is contrary to the existing literature (Anderson & van Wincoop, 2011; Baier & Bergstrand, 2007) and contrary to the original model (Tinbergen, 1962). For the regression with the countries with the FTA set after 2004, the variable is 0.0313. The variable lGDPEU is in both regressions significantly positive and this is in line with the existing literature and the original model.
The variable lGDPCAP is significantly positive for the countries with the FTA set before the enlargement, but not statisically significant different from zero for the countries with the FTA set after the enlargement. The variable lGDPCAPEU is significantly negative for both regressions.
19
Discussion & Conclusion Discussion
The dummy variable FTAijt is positive, but not statistically significant different than zero for the countries with the FTA’s set before the enlargement. This is remarkable, since we expect FTA’s to have a positive effect on the bilateral trade of the partner countries. One reason for this is that the countries included in the dataset had FTA’s set in the same time period with other countries as well. Mexico set an FTA with Chile in 1999, Armenia with Kazachstan in 2001 and Israel with Turkey in 1998 (Baier & Bergstrand, 2007). This may result in higher overall exports for these countries, but not particularly in higher exports towards the EU. It is possible that the time-fixed effects did not fully account for these events.
The dummy variable FTAijt is significantly positive for the countries with the FTA set after the enlargement. This implies that the FTA’s set in Korea, Albania, Serbia and Georgia contributed to a rise in exports. This is in line with the existing literature (Yang & Martínez-Zarzoso, 2013; Soete & van Hove, 2017) . This thesis aimed to analyze whether there is a difference between the effects of FTA’s on trade before and after the enlargement of the EU in 2004. It is consistent with the alternative hypothesis of this thesis that when the Free Trade Area becomes larger, the positive effects on their bilateral trade are increasing as well. More economies of scale, better use of comparative advantages and higher exports due to lower tariffs, will enhance the bilateral exports (Krugman et al, 2009). The results show that the effect of the FTA’s of the EU on trade to has risen after the enlargement in 2004.
The variable of lGDP is significantly negative in the regression for the countries with the FTA set before 2004. This implies that an increase in GDP will lead to a decrease in the bilateral exports. The original model and existing literature expects that an increase in GDP will lead to an increase in bilateral exports. Namely, a country with a bigger economic size is expected to trade more. One explanation for this is the burst of the dot-com bubble in March, 2000. Even though the exports kept increasing for the countries, in the early 2000s the GDP was not increasing as fast as first. It is possible that the time-fixed effects did not fully account for this event. The same holds for the countries with the FTA set after 2004. This variable is not significantly positive either.
The variable lGDPCAP is used in the model to control the lGDP variable as described in chapter 2 by the populations of the country. The variable is according to the original model and most existing literature negative. In this thesis, the variable is negatively related to the variable lGDP, so the variable lGDPCAP controls for the variable lGDP. Due to the
20
remarkable outcome in the regression with the countries with the FTA set before 2004, the variable lGDPCAP has a deviating outcome as well. The variable lGDPCAPEU is
significantly negative as in line with the original model and existing literature (Endoh, 2000).
The impact of fixed effects
The specification of the gravity model has an impact on its results, so this thesis tries to contribute to the academic debate on how to use the gravity model as well. Column 2 and 4 in table 3 show the panel data regressions without fixed effects. The regression is the same as the regression of the baseline model, except now country-pair fixed effects and time fixed effects are left out of the regression. The impact of the fixed effects can be determined by analyzing the differences of the regressions. For the countries with the FTA set after 2004, all variables except the variable FTA and lGDP are the opposite of what the regression found with fixed effects. lGDP and GDPCAP will turn significant. Meanwhile lGDPEU,
lGDPCAPEU and the constant variable will turn insignificant. It is in line with Glick & Rose (2003) that a misspecification of the regression model will lead to significant different
outcomes. It is in line with Anderson & van Wincoop (2011) that leaving out fixed effects in the regression will lead to biased results. The model without fixed effects tells that the FTA’s had a significant effect on trade. Moreover, the contribution of the FTA’s on trade will be 141.38%1. This unlikely high outcome can be explained by the exclusion of time fixed effects. The exports are namely for all four countries increasing over time. This is not only applicable to the set of the FTA’s, but due to globalization and the rise connectivity of world trade as well (Storper, 1992). The exports were already rising before the FTA was set. The fixed effects are able to account for these external factors.
For the regression with the countries with the FTA set before 2004, the fixed effects had a significant effect on the regression as well. All variables are significant in the
regression with fixed effects except for the dummy variable FTAijt. When excluding the fixed effects, all variables turn insignificant except the variables FTAijt and lGDPCAP. The dummy variable FTAijt turned to a value of -0.8599. This implies that FTA’s set in Armenia, Mexico, Morocco and Israel had a negative effect on trade of 136.29%2. This leads to a negative number of trade, so this is impossible to be true.
The R-squared indicates the proportion of the variance of the dependent variable that is explained by the independent variables (Stock & Watson, 2007). A higher R-squared
1𝑒^(0.8812) − 1 = 1.4138 2𝑒^(0.8599) − 1 = 1.3629
21
results a higher goodness of fit. The bottom row of table 3 shows that the model with fixed effects explains the variance of the dependent variable better than the model without fixed effects.
The results show that it is important to include fixed effects in the regression in order to get unbiased results. It solves the endogeneity issue for the most part. The problem that unusually high trade flows lead to FTA’s and not vice versa can lead to biased results is now accounted for (Micco et al., 2007). Moreover, the fixed effects are necessary to capture the time-specific and country-specific events. The fixed effects allow the dummy variable FTA to isolate the effect of the FTA’s on trade. The countries’ heterogeneity is omitted, when the country-pair fixed effects and time fixed effects are left out the regression. The gravity model cannot be augmented with unlimited variables to capture every specific event the fixed effects can capture (Baier & Bergstrand, 2007).
Conclusion
This thesis aimed to analyze whether there is a difference between the effects of FTA’s on trade before and after the enlargement of the EU in 2004. This was done by utilizing a theoretically-motivated gravity model using panel data with time fixed effects and cross-country fixed effects. Two regressions are done in this thesis. The first regression aims to analyze the effect of FTA’s on trade for countries with an FTA set before 2004. The second regression aims to do the same, but for countries with an FTA set after 2004. The results indicate that for the countries where the FTA was set before 2004, the effect of the FTA on trade was not significant. For the countries where the FTA was set after 2004, the effect of the FTA on trade was found significant. The FTA’s set in Korea, Albania, Serbia and Georgia contributed to a rise in exports of 12.55%. This thesis concludes that the enlargement of the EU had a positive effect for its FTA’s on trade, by comparing these results.
The use of time fixed effects and cross-country effects is of high importance, when using the gravity model. This thesis contributes to the academic debate on how to use the gravity model. The use of fixed effects has an important impact on its results, as in this thesis shown. The fixed effects allow the dummy variable FTAijt to isolate the effect of the FTA’s on trade (Baier & Bergstrand, 2007). Moreover, it accounts for most of the endogeneity bias (Micco et al., 2007).
Further research can be done on this subject by enlarging the dataset and by doing a regression. Extending the dataset will enhance the generalizability of this thesis. The IV-regression will account for the whole endogeneity bias. At last, this thesis is limited in
22
examining the economic rationale behind some of the findings. Further research could investigate this further and give an exact interpretation of the results.
Overall, the results point out that the enlargement of the EU in 2004 had an effect on the FTA’s of the EU. The effects of the FTA’s of the EU enhanced with 12.55%, due to the enlargement of the EU. More economies of scale, better use of comparative advantages and higher exports due to lower tariffs with more countries contributed positively on the trade effect of the FTA’s of the EU after the enlargement in 2004.
23
References
Abrams, R. K. (1980). International Trade Flows under Flexible Exchange Rates. Economic Review, 65 (3), 3–10.
Aitken, N. (1973). The Effect of the EEC and EFTA on European Trade: A Temporal Cross-Section Analysis. The American Economic Review, 63(5), 881-892.
Anderson, J. E. (1979). A Theoretical Foundation for the Gravity Equation. The American Economic Review, 69(1), 106–116.
Anderson, J. E. (2011). The Gravity Model. Annual Review of Economics, 3(1), 133-160. Appleyard, D. R., & Field, A.J. (2010). International Economics. McGraw-Hill Education.
7thedition
Baier, S. L., Bergstrand, J. H., (2007). Do Free Trade Agreements Actually Increase Members’ International Trade? Journal of International Economics. 71, 72-95. Balassa, B. (1961). The Theory of Economic Integration. Homewood, Ill.: Richard D. Irwin,
Inc.
Balassa, B. (1974). Trade Creation and Trade Diversion in the European Common Market: An Appraisal of the Evidence. Manchester School of Economic and Social Studies, 42(2), 93–135.
Bayoumi T., & Eichengreen B., (1997 ). Ever closer to heaven? An optimum-currency-area index for European countries, European Economic Review. 41(3–5), 761-770.
Belke, A., & Spies, J. (2008). Enlarging the EMU to the East what Effect on Trade. Empirica, 35(4), 369–389.
Bergstrand, J. H. (1985). The Gravity Equation in International Trade: Some Microeconomic Foundations and Empirical Evidence. The Review of Economics and Statistics, 474– 481.
Breuss, F. (2009). An Evaluation of the EU’ Fifth Enlargement With special focus on Bulgaria and Romania. Economic Papers, 361.
Cavallo, A.E., & Frankel, J.A. (2008) Does openness to trade make countries more
vulnerable to sudden stops, or less? Using gravity to establish causality, Journal of International Money and Finance, 27(8), 1430-1452.
Cheng I., & Wall H. J., (2005). Controlling for Heterogeneity in Gravity Models of Trade and Integration. Federal Reserve Bank of St. Louis Review, 87(1), 49-63
Deardorff A. V., (1998). Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World?, The Regionalization of the World Economy, 7-32.
Dornbusch R., Fischer S. , & Samuelson P.A. (1977). Comparative Advantage, Trade, and Payments in a Ricardian Model with a Continuum of Goods, The American Economic Review, 67(5), 823-839.
24
Endoh, M. (1999) The transition of postwar Asia-Pacific trade relations, Journal of Asian Economics, 10(4), 571-589.
European Commission. Treaty on European Union. (1992, February 7). Retrieved January 8, 2018 from https://europa.eu/european-union/sites/europaeu/files/docs/
body/treaty_on_european_union_en.pdf
European Commission. Enlarging the European Union: from 15 to 25, what does it mean for us?. (2005, June 5). Retrieved January 5, 2018 from
https://ec.europa.eu/neighbourhood-enlargement/sites/near/files/
archives/pdf/press_corner/publications/key_issues_fr om15_to_25_en.pdf European Commission. Negotiations and agreements. (2017, november 18). Retrieved
November 20, 2017, from http://ec.europa.eu/trade/policy/countries-and-regions/negotiations-and- agreements/#_being-negotiated
European Commission. What is the Economic and Monetary Union? (EMU). (2017,
December 18). Retrieved January 14, 2018, from https://ec.europa.eu/info/business- economy-euro/economic-and-fiscal-policy-coordination/economic-and-monetary-union/what-economic-and-monetary-union-emu_en
Frankel, J. A., Stein, E., & Wei, S.-J. (1997). Regional Trading Blocs in the World Economic System. Peterson Institute.
Hausman A.J., & Taylor W.E., (1981). A generalized specification test, Economics Letters, 8(3), 239-245
Kiesraad (2016). Uitslag referendum Associatieovereenkomst met Oekraïne, April 4th 2016,
https://www.kiesraad.nl/actueel/nieuws/2016/04/12/uitslag-referendum-associatieovereenkomst-met-oekraine
Koshal, R. (1972). Economies of Scale. Journal of Transport Economics and Policy, 6(2), 147-153.
Krugman, P. R., Obstfeld, M. (2009). International economics: Theory and policy. Boston: Pearson Addison-Wesley.
Matyas, L. (1997). Proper Econometric Specification of the Gravity Model. World Economy, 20(3), 363-368.
McCallum, J. (1995). National Borders Matter: Canada-US Regional Trade Patterns. The American Economic Review, 615–623.
Micco, A., Stein, E., & Ordõnez, G. (2003). The Currency Union Effect on Trade: Early Evidence from EMU. Economic Policy, 18 (37), 315–356.
Rose, A. K. (2004). "Do We Really Know That the WTO Increases Trade?" American Economic Review , 98-114.
Soete, S. & Van Hove, J. (2017) Dissecting the Trade Effects of Europe's Economic Integration Agreements, Journal of Economic Integration, 32(1), 193-243.
25
Stock, J. H. & Watson, M. (2007). Introduction to Econometrics. Pearson Education, Inc. 2nd edition.
Storper, M. (1992). The Limits to Globalization: Technology Districts and International Trade, Economic Geography , 68(1), 60-93.
Tinbergen, J. (1962). Shaping the World Economy. Twentieth Century Fund. Viner, J. (2014). The Customs Union Issue. Oxford University Press.
Wadhva, C. D. (1971) The theory of Customs Union and Free Trade Areas among developing countries for Balance of Payments objectives: an explanatory approach. Indian Economic Journal,19(2), 168-186.
Wood, J. C. (1991). David Ricardo: Critical Assessments. Taylor & Francis.
Yang, S., & Martíınez-Zarzoso, I. (2013). A Panel Data Analysis of Trade Creation and Trade Diversion Effects: the Case of ASEAN-China Free Trade Area (ACFTA). Discussion Papers(224).
